Tensorflow.js tf.variableGrads() Function
Last Updated :
30 May, 2021
Tensorflow.js is an open-source library that is developed by Google for running machine learning models as well as deep learning neural networks in the browser or node environment.
The .variableGrads() function is used to calculate as well as return the gradient of f(x) in comparison to the stated list of manageable variables that are presented by the parameter varList. Moreover, if the list is not given then by default it is all the manageable variables.
Syntax:
tf.variableGrads(f, varList?)
Parameters:
- f: It is the stated function which is to be executed. Where, f() must return a scalar. It is of type (() => tf.Scalar).
- varList: It is the stated list of variables that are used to calculate the gradients in comparison to and by default it is all the manageable variables. It is of type tf.Variable[].
Return Value: It returns value which is of type tf.Scalar and it also returns grads which is of type {[name: string]: tf.Tensor}.
Example 1:
Javascript
import * as tf from "@tensorflow/tfjs"
const p = tf.variable(tf.tensor1d([9, 6]));
const q = tf.variable(tf.tensor1d([7, 8]));
const r = tf.tensor1d([3, 4]);
const fn = () => p.add(r.square()).mul(q.add(r)).sum();
const {val, grads} = tf.variableGrads(fn);
Object.keys(grads).forEach(
variable_Name => grads[variable_Name].print());
|
Output:
Tensor
[10, 12]
Tensor
[18, 22]
Example 2:
Javascript
import * as tf from "@tensorflow/tfjs"
const p = tf.variable(tf.tensor1d([3.1, 5.2]));
const q = tf.variable(tf.tensor1d([4.4, 6.7]));
const r = tf.tensor1d([7.1, 3.2]);
const {val, grads} = tf.variableGrads(
() => p.add(r.square()).mul(q.add(r)).sum());
Object.keys(grads).forEach(
variable_Name => grads[variable_Name].print());
|
Output:
Tensor
[11.5, 9.8999996]
Tensor
[53.5099983, 15.4400005]
Reference: https://js.tensorflow.org/api/latest/#variableGrads
Please Login to comment...